捕获服务器操作失败与遥测数据
将操作错误和性能指标连同用户上下文报告到可观测性后端。
捕获服务器操作失败与遥测数据 是 CoddyKit 上的免费 Next.js 15 Fullstack (App Router + Server Actions) 课时。 这是第 4 节课，共 4 节。 你可以在下方免费阅读本课时的完整内容 — 然后在浏览器中使用内置代码编辑器和全天候 AI 导师进行实践。 这是 Next.js 15 Fullstack (App Router + Server Actions) 学习路径的一部分，你的进度在网页和 CoddyKit 应用中同步。 Next.js 15 Fullstack (App Router + Server Actions) 课程共包含 4 节课。
本课时的部分内容尚未翻译，以英文显示。
Why Server Action Failures Need Special Treatment
Server Actions in Next.js 15 run on the server, outside the browser's error boundary. When a Server Action throws, the client receives a generic error — no stack trace, no context, no telemetry.
This means standard try/catch in the action is necessary but not sufficient. You also need to:
- Capture the error with full server-side context (user ID, action name, request metadata)
- Forward structured telemetry to an observability backend (e.g. OpenTelemetry, Sentry, Datadog)
- Return a safe, serialisable error payload to the client — never expose raw stack traces
In this lesson you will build a complete, production-grade error-and-metrics pipeline around Server Actions.
Structuring a Safe Action Result Type
Before adding telemetry, agree on the shape every Server Action returns. A discriminated union makes error handling exhaustive on the client and prevents accidental data leaks.
Define this once in a shared types file and import it everywhere:
// lib/action-result.ts
export type ActionSuccess<T> = {
ok: true;
data: T;
durationMs: number;
};
export type ActionFailure = {
ok: false;
code: string; // machine-readable, e.g. 'VALIDATION_ERROR'
message: string; // human-readable, safe to show
durationMs: number;
};
export type ActionResult<T> = ActionSuccess<T> | ActionFailure;
// Helper constructors
export function ok<T>(data: T, durationMs: number): ActionSuccess<T> {
return { ok: true, data, durationMs };
}
export function fail(code: string, message: string, durationMs: number): ActionFailure {
return { ok: false, code, message, durationMs };
}Instrumenting Duration at the Action Level
Performance metrics begin with measuring how long each action takes. Wrap every action body with a performance.now() call so duration is always available for telemetry — even when the action fails.
The pattern below ensures durationMs is captured in both success and error paths:
// app/actions/create-post.ts
'use server';
import { ok, fail, type ActionResult } from '@/lib/action-result';
import { db } from '@/lib/db';
export async function createPost(
_prev: ActionResult<{ id: string }> | null,
formData: FormData
): Promise<ActionResult<{ id: string }>> {
const start = performance.now();
try {
const title = formData.get('title') as string;
if (!title?.trim()) {
return fail('VALIDATION_ERROR', 'Title is required', performance.now() - start);
}
const post = await db.post.create({ data: { title } });
return ok({ id: post.id }, performance.now() - start);
} catch (err) {
return fail('INTERNAL_ERROR', 'Failed to create post', performance.now() - start);
}
}Capturing User Context in Server Actions
Observability is only useful when you know who experienced an error. Server Actions have access to the same server context as Route Handlers, so you can read the session before running business logic.
Use Next-Auth (or any session library) to attach user context to every telemetry event:
// lib/auth-context.ts
import { auth } from '@/auth'; // next-auth v5
export type UserContext = {
userId: string | null;
email: string | null;
role: string | null;
};
export async function getUserContext(): Promise<UserContext> {
const session = await auth();
return {
userId: session?.user?.id ?? null,
email: session?.user?.email ?? null,
role: (session?.user as { role?: string })?.role ?? null,
};
}Building a Telemetry Reporter
Centralise all telemetry calls in a single reporter module. This keeps observability logic out of business code and makes it easy to swap providers (Sentry, Datadog, OpenTelemetry) without touching action files.
The reporter records:
- Errors — with user context, action name, and serialised error
- Metrics — action name, outcome, and duration for latency percentiles
// lib/telemetry.ts
import type { UserContext } from './auth-context';
export type TelemetryEvent = {
action: string;
ok: boolean;
durationMs: number;
code?: string;
userId?: string | null;
email?: string | null;
};
export async function reportEvent(event: TelemetryEvent): Promise<void> {
// Replace with your actual provider SDK call, e.g.:
// Sentry.captureMessage(...) or
// otelMeter.histogram(...)
if (process.env.NODE_ENV === 'production') {
await fetch(process.env.OBSERVABILITY_INGEST_URL!, {
method: 'POST',
headers: { 'Content-Type': 'application/json',
'X-Api-Key': process.env.OBSERVABILITY_API_KEY! },
body: JSON.stringify({ ...event, timestamp: Date.now() }),
});
} else {
console.info('[telemetry]', event);
}
}
export async function reportError(
action: string,
err: unknown,
ctx: UserContext,
durationMs: number
): Promise<void> {
const message = err instanceof Error ? err.message : String(err);
console.error(`[action:${action}] ${message}`, { userId: ctx.userId });
await reportEvent({
action, ok: false, durationMs,
code: 'INTERNAL_ERROR',
userId: ctx.userId,
email: ctx.email,
});
}Composing the Full Instrumented Action
Now combine user context, duration measurement, and telemetry reporting into one coherent Server Action. The key rule is: always report before returning — even on the happy path, so you have latency data for successful calls too.
// app/actions/create-post.ts (final version)
'use server';
import { ok, fail, type ActionResult } from '@/lib/action-result';
import { getUserContext } from '@/lib/auth-context';
import { reportEvent, reportError } from '@/lib/telemetry';
import { db } from '@/lib/db';
export async function createPost(
_prev: ActionResult<{ id: string }> | null,
formData: FormData
): Promise<ActionResult<{ id: string }>> {
const start = performance.now();
const ctx = await getUserContext();
try {
const title = (formData.get('title') as string)?.trim();
if (!title) {
const dur = performance.now() - start;
await reportEvent({ action: 'createPost', ok: false,
code: 'VALIDATION_ERROR', durationMs: dur,
userId: ctx.userId });
return fail('VALIDATION_ERROR', 'Title is required', dur);
}
const post = await db.post.create({ data: { title, authorId: ctx.userId! } });
const dur = performance.now() - start;
await reportEvent({ action: 'createPost', ok: true,
durationMs: dur, userId: ctx.userId });
return ok({ id: post.id }, dur);
} catch (err) {
const dur = performance.now() - start;
await reportError('createPost', err, ctx, dur);
return fail('INTERNAL_ERROR', 'Failed to create post', dur);
}
}Extracting a withAction Higher-Order Wrapper
Duplicating context-fetch and telemetry calls in every action creates noise and inconsistency. Extract a higher-order wrapper, withAction, that handles the boilerplate automatically.
The wrapper:
- Reads user context once at the start
- Measures duration around the inner function
- Reports success and failure telemetry
- Translates unknown thrown errors into safe
ActionFailurevalues
// lib/with-action.ts
import { ok, fail, type ActionResult } from './action-result';
import { getUserContext, type UserContext } from './auth-context';
import { reportEvent, reportError } from './telemetry';
type InnerFn<TInput, TOutput> = (
input: TInput,
ctx: UserContext
) => Promise<ActionResult<TOutput>>;
export function withAction<TInput, TOutput>(
name: string,
inner: InnerFn<TInput, TOutput>
) {
return async (input: TInput): Promise<ActionResult<TOutput>> => {
const start = performance.now();
const ctx = await getUserContext();
try {
const result = await inner(input, ctx);
const dur = performance.now() - start;
await reportEvent({
action: name, ok: result.ok,
code: result.ok ? undefined : (result as { code: string }).code,
durationMs: dur, userId: ctx.userId,
});
return result;
} catch (err) {
const dur = performance.now() - start;
await reportError(name, err, ctx, dur);
return fail('INTERNAL_ERROR', 'An unexpected error occurred', dur);
}
};
}Using withAction in a Real Server Action
With withAction in place, individual action files become concise and focused purely on business logic. All observability wiring is invisible to the developer writing the action.
// app/actions/delete-post.ts
'use server';
import { ok, fail } from '@/lib/action-result';
import { withAction } from '@/lib/with-action';
import { db } from '@/lib/db';
export const deletePost = withAction(
'deletePost',
async ({ postId }: { postId: string }, ctx) => {
if (!ctx.userId) {
return fail('UNAUTHENTICATED', 'You must be logged in', 0);
}
const post = await db.post.findUnique({ where: { id: postId } });
if (!post) {
return fail('NOT_FOUND', 'Post not found', 0);
}
if (post.authorId !== ctx.userId) {
return fail('FORBIDDEN', 'You do not own this post', 0);
}
await db.post.delete({ where: { id: postId } });
return ok({ deleted: true }, 0); // duration injected by wrapper
}
);
// Usage in a Client Component:
// const result = await deletePost({ postId: '...' });
// if (!result.ok) toast.error(result.message);Connecting to OpenTelemetry Traces
Metric events are useful, but distributed traces let you see exactly where time was spent across DB queries, external API calls, and middleware. Next.js 15 has built-in OpenTelemetry support via the instrumentation.ts file.
Register your OTel SDK once, then create spans inside actions to complement the automatic HTTP spans:
// instrumentation.ts (Next.js 15 — runs once on server start)
export async function register() {
if (process.env.NEXT_RUNTIME === 'nodejs') {
const { NodeSDK } = await import('@opentelemetry/sdk-node');
const { OTLPTraceExporter } = await import('@opentelemetry/exporter-trace-otlp-http');
const { Resource } = await import('@opentelemetry/resources');
const { SEMRESATTRS_SERVICE_NAME } = await import('@opentelemetry/semantic-conventions');
const sdk = new NodeSDK({
resource: new Resource({
[SEMRESATTRS_SERVICE_NAME]: 'my-nextjs-app',
}),
traceExporter: new OTLPTraceExporter({
url: process.env.OTEL_EXPORTER_OTLP_ENDPOINT,
}),
});
sdk.start();
}
}Adding Custom Spans Inside withAction
Once OTel is registered, you can enrich traces with custom spans directly inside the withAction wrapper. This creates a parent span for the entire action, which automatically parents any child spans created by DB clients or fetch calls during the action's execution.
// lib/with-action.ts (OTel-enhanced version)
import { trace, SpanStatusCode } from '@opentelemetry/api';
import { ok, fail, type ActionResult } from './action-result';
import { getUserContext, type UserContext } from './auth-context';
import { reportEvent, reportError } from './telemetry';
const tracer = trace.getTracer('server-actions');
type InnerFn<I, O> = (input: I, ctx: UserContext) => Promise<ActionResult<O>>;
export function withAction<I, O>(name: string, inner: InnerFn<I, O>) {
return async (input: I): Promise<ActionResult<O>> => {
return tracer.startActiveSpan(`action.${name}`, async (span) => {
const start = performance.now();
const ctx = await getUserContext();
span.setAttributes({
'action.name': name,
'user.id': ctx.userId ?? 'anonymous',
'user.email': ctx.email ?? '',
});
try {
const result = await inner(input, ctx);
const dur = performance.now() - start;
span.setAttributes({ 'action.ok': result.ok, 'action.duration_ms': dur });
span.setStatus({ code: result.ok ? SpanStatusCode.OK : SpanStatusCode.ERROR });
span.end();
await reportEvent({ action: name, ok: result.ok, durationMs: dur, userId: ctx.userId });
return result;
} catch (err) {
const dur = performance.now() - start;
span.recordException(err as Error);
span.setStatus({ code: SpanStatusCode.ERROR });
span.end();
await reportError(name, err, ctx, dur);
return fail('INTERNAL_ERROR', 'An unexpected error occurred', dur);
}
});
};
}Surfacing Action Errors in a Client Error Boundary
Even with perfect server-side telemetry, the client still needs to handle failures gracefully. Use useActionState to relay the ActionResult back to the UI, and pair it with a React Error Boundary for unexpected render errors:
useActionStatepropagates theActionFailurepayload so you can show inline messages- A top-level
error.tsxin the App Router catches errors that escape action boundaries (e.g. during streaming) - Log client-side boundary triggers back to the same telemetry pipeline via a dedicated API route to keep a unified error view
This completes the loop: errors are captured on the server with full context and surfaced on the client with safe, user-friendly messages.
Knowledge Check: Telemetry in Server Actions
Test your understanding of the key design decisions in this lesson.
Lesson Recap: Server Action Telemetry Pipeline
In this lesson you built a complete observability pipeline for Next.js 15 Server Actions:
- ActionResult discriminated union — a consistent, serialisable return type that distinguishes success from failure without exposing server internals to the client
- Duration measurement —
performance.now()wrapping every action body, captured on both success and error paths - User context — reading the session once at the start of each action so every telemetry event carries
userIdandemail - Centralised telemetry reporter — a single module that forwards structured events to your observability backend, keeping business logic clean
- withAction wrapper — a higher-order function that composes context, duration, and telemetry automatically so individual actions stay focused on business logic
- OpenTelemetry integration — registering the OTel SDK in
instrumentation.tsand creating parent spans insidewithActionso DB and external API calls appear as child spans in distributed traces - Client error surfacing — using
useActionStateanderror.tsxto relay safe failure messages to users while keeping raw errors server-side
Together these layers give you full visibility into every action invocation — who triggered it, whether it succeeded, how long it took, and exactly where it failed.
用 AI 导师学习 TypeScript — 免费
在浏览器中编写并运行真实代码，获得全天候 AI 导师的即时帮助，并在网页或应用中继续学习。
- 课程
- 22
- 课程
- 88
常见问题解答
「捕获服务器操作失败与遥测数据」课时是免费的吗？
是的 — 「捕获服务器操作失败与遥测数据」的完整文本可在网页上免费阅读。要进行交互式练习（内置代码编辑器和全天候 AI 导师）并解锁 Next.js 15 Fullstack (App Router + Server Actions) 课程的其余内容，请升级到 CoddyKit PRO。 Next.js 15 Fullstack (App Router + Server Actions) 课程共包含 4 节课。
「捕获服务器操作失败与遥测数据」这节课中我会学到什么？
将操作错误和性能指标连同用户上下文报告到可观测性后端。 你通过在浏览器中直接运行的动手代码来练习 Next.js 15 Fullstack (App Router + Server Actions)，全天候 AI 导师会在你学习这节课的过程中回答你的问题。
学习 Next.js 15 Fullstack (App Router + Server Actions) 需要有经验吗？
无需任何先前经验。CoddyKit 上的 Next.js 15 Fullstack (App Router + Server Actions) 课程适合初学者到高级学习者，你可以从这里开始或从头开始，按照自己的节奏学习。 这是第 4 节课，共 4 节。
「捕获服务器操作失败与遥测数据」课时需要多长时间？
大多数 CoddyKit 课程大约需要 5–10 分钟。每节课都很精短且互动，所以你能稳步进步，并在网页和应用中从离开的地方继续。
我能在这节 Next.js 15 Fullstack (App Router + Server Actions) 课中编写并运行代码吗？
能。每节 Next.js 15 Fullstack (App Router + Server Actions) 课都包含内置代码编辑器，你可以在浏览器中直接编写并运行真实代码，并获得即时 AI 反馈 — 无需本地设置。